CentOS - Feb 2016 - [Bulk] Re: safest way to mount iscsi loopback..

On 2/11/2016 5:14 AM, lejeczek wrote:
> nobody does use iscsi loopback over an lvm? 
I'm not sure what 'iscsi loopback' even means.    iSCSI is used to
mount
a virtual block device hosted on another system (initiator mode) or to 
share a virtual block device (target mode), while loopback is used to 
mount a local file as a device, such as an .iso image of an optical disc.

can you explain in a little more detail what you're trying to do ?



-- 
john r pierce, recycling bits in santa cruz

On 11/02/16 20:20, John R Pierce wrote:
> On 2/11/2016 5:14 AM, lejeczek wrote:
>> nobody does use iscsi loopback over an lvm? 
>
> I'm not sure what 'iscsi loopback' even means.    iSCSI is 
> used to mount a virtual block device hosted on another 
> system (initiator mode) or to share a virtual block device 
> (target mode), while loopback is used to mount a local 
> file as a device, such as an .iso image of an optical disc.
>
> can you explain in a little more detail what you're trying 
> to do ?
>
>
>
whatever devices you have in your backstores (LIO 
implementation is the default one I believe) and then 
naturally in your targets, etc.. - on the same local system 
you can loop them back = LIO presents them again to the 
kernel as local scsi devices.
I'm thinking, maybe multipath should be involved/deployed here?
I can mount such a loopback iscsi target but I cannot use 
FS's uuid because - in my case first UUID comes from LVM's 
lv and then the same UUID comes via iscsi loopback - which 
makes sense for it's the same one filesystem.

On Mon, 15 Feb 2016, lejeczek wrote:
> whatever devices you have in your backstores (LIO implementation is the 
> default one I believe) and then naturally in your targets, etc.. - on the 
> same local system you can loop them back = LIO presents them again to the 
> kernel as local scsi devices.
> I'm thinking, maybe multipath should be involved/deployed here?
> I can mount such a loopback iscsi target but I cannot use FS's uuid
because -
> in my case first UUID comes from LVM's lv and then the same UUID comes
via
> iscsi loopback - which makes sense for it's the same one filesystem.
Why would you be trying to do some loopback fun when you could just mount the
underlying block device directly?  Why not avoid loopback and mount the LV?

jh